Enhancing convergence efficiency of self-propelled agents using direction preference
Introduction
Collective behavior is very common in nature, such as bird flocks [1], [2], fish schools [3], [4], [5], [6], insect swarms [7], [8] and bacterial colonies [9], [10]. This amazing natural phenomenon has attracted many physicists, biologists and control scientists to study. They explore the nature of collective behavior through observation, simulation, and mathematical proof [11].
In 1987, Reynolds simulated the first collective motion through the computer program [12]. In his model, three rules of collective motion were proposed: aggregation, synchronization and separation. These three principles provide direction for the research of collective motion [13], [14], [15], [16], [17], [18], [19], [20], [21]. In terms of cluster synchronization, the most famous model was proposed by Vicsek [19], now well known as the Vicsek model (VM). In the VM, all agents are distributed in a periodic square environment randomly and have the same fixed speed and communication radius. In direction updating, each agent takes the average direction of all its neighbors (include itself) within the communication radius as its next step direction. The mechanism of the VM is very simple, but it can well simulate the phase transition process from disorder to order in nature. Because it is simple and efficient, the VM is considered as the basic model for collective motion. The VM has been further studied by other researchers [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33]. For instance, Ihle uses the quantitative kinetic theory to outline the used theory of the VM [24]. Romensky et al. believe that agents prefer to synchronize with their neighbors who are in a similar direction and focus their research on the effect of velocity restriction on the phase transition of the VM [34]. Piwowarczyk et al. study the influence of sensorial delay on clustering and swarming and find that short delays can enhance the emergence of clusters and swarms in VM [23]. Some researchers add the view angle to the VM and study the effect of view angle on collective motion [25], [26], [27], [28]. Li et al. modify the constant-speed VM into a variable-speed model, the studies show that the modified model is easier to achieve direction synchronization than the original VM [33].
The original VM is a homogeneous model. When agents process information, they treat their neighbors equally. From the biological point of view, this is very unreasonable, because biological individuals are inevitably affected by subjective and objective factors when processing information. For example, in the study of the flock of pigeons, it was found that the influence between individuals is not equivalent [35]. In a crowd, individuals prefer to synchronize with people they know. Many studies on non-homogeneous models also show that non-homogeneous models can make the network easier to synchronize [36], [37] and improve the cooperation between individuals [38]. Inspired by this idea, many researchers have proposed weight models from different perspectives. Yang et al. set the weight of neighbors based on the geographical distance of the agents and their neighbors [30]. Gao et al. judge the weight of neighbors by the density of agents around the neighbors [31]. Zou et al. consider that the weight of neighbors should be determined by the angle between the position of neighbor and direction of agents [32].
In this paper, we propose a direction preference model in which agents prefer to synchronize with their neighbors moving in a certain direction. The center agent judges the influence value from its neighbor according to the direction angle between them. We assume that there exists a preference direction angle . When the direction angle approaches , the neighbor has greater influence on the center agent. In this paper, we concentrate our research on the influence of different on convergence time; whereas, Romensky et al. focus their research on the effect of the velocity restriction on the phase transition of the VM at the case [34]. The parameter is used to adjust the effect of direction preference. The larger the value of , the weaker the effect of direction preference. When , the direction preference model will become the original Vicsek model. The effect of the direction preference model on the collective motion is investigated via a large number of simulation experiments under different conditions of restriction angle, absolute velocity, and combination parameter consisting of communication radius and density. Simulation results show that compared to the original VM, in noise-free environment the direction preference model with optimal can accelerate the synchronization speed; in noise environment the direction preference model with be in has stronger robustness and the optimal value of is altered.
Section snippets
The Vicsek model
In Vicsek model, a group of agents move in a square-shape cell with a fixed velocity . Initially, all agents are distributed in the square zone with periodic boundary randomly and the direction of agents is distributed in randomly. In the system, each agent only knows local information, which means they can only get information from neighbors within the radius . All agents update their directions with the same rule. At each time step, agents update their directions by the
The modified model
Inspired by the cluster non-homogeneity, we propose a direction preference model (DPM) based on the original VM. In the DPM, it is assumed that the importance of each agent to other agents is different. When updating the direction, each agent evaluates the influence value from its neighbor according to the direction angle between them. Therefore, the direction angle is a very important evaluation criterion, which is defined as the angle between the direction of agent (center
Simulation and discussion
In order to investigate the effect of the proposed direction preference model on flocking synchronization, we compare the modified model (DPM) with the original model (VM) under different conditions through computer simulation. In these simulations, agents live in square zone with periodic boundary conditions. Due to the limited communication capacity, each agent can only influence other agents within the communication radius . In general, all agents has the same absolute velocity at
Conclusions
Inspired by cluster diversity and non-homogeneity, some researchers have investigated different non-homogeneous models based on VM [30], [31], [32]. In this paper, we proposed a new non-homogeneous model. In this model, agents prefer to synchronize with their neighbors moving in a preference direction. In addition, restriction angle is considered in this study, which allows our findings to be better applied in real world. Both noise-free and noise cases are considered in this paper. In
CRediT authorship contribution statement
Yu-Rong Chen: Conceptualization, Methodology, Software, Writing – original draft. Xian-Xia Zhang: Investigation, Formal analysis, Writing – review & editing, Supervision, Funding acquisition. Yin-Sheng Yu: Simulation, Experimental analysis. Shi-Wei Ma: Resources, Project administration. Banghua Yang: Visualization, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by the project from the National Natural Science Foundation of China under Grant 62073210 and Grant 61976133
References (39)
- et al.
Context-dependent group size choice in fish
J. Theoret. Biol.
(2004) - et al.
Behavior pattern (innate action) of individuals in fish schools generating efficient collective evasion from predation
J. Theoret. Biol.
(2005) - et al.
Collective motion and cannibalism in locust migratory bands
Curr. Biol.
(2008) - et al.
Collective motion
Phys. Rep.-Rev. Sec. Phys. Lett.
(2012) - et al.
Simulating predator attacks on schools: Evolving composite tactics
Ecol. Model.
(2015) - et al.
Collective memory and spatial sorting in animal groups
J. Theoret. Biol.
(2002) - et al.
Reducing convergence times of self-propelled swarms via modified nearest neighbor rulesn
Physica A
(2012) - et al.
Improving the synchronization speed of self-propelled particles with restricted vision via changing the line of sight randomly
Nonlinear Dynam.
(2017) - et al.
A modified vicsek model for self-propelled agents with exponential neighbor weight and restricted visual field
Physica A
(2013) - et al.
Diversity of neighborhoods promotes cooperation in evolutionary social dilemmas
Physica A
(2017)
Accelerating consensus of self-driven swarm via adaptive speed
Physica A
Statistical mechanics of complex networks
Rev. Modern Phys.
Spatiotemporal dynamics of skeins of wild geese
EPL
Consequences of variation in predator attack for the evolution of the selfish herd
Evol. Ecol.
Predator confusion is sufficient to evolve swarming behaviour
J. R. Soc.
From disorder to order in marching locusts
Science
Concentration dependence of the collective dynamics of swimming bacteria
Phys. Rev. Lett.
Periodic reversal of direction allows Myxobacteria to swarm
Proc. Natl. Acad. Sci. USA
Detecting communities by the core-vertex and intimate degree in complex networks
Comput. Graph.
Cited by (2)
Enhancing synchronization of self-propelled particles via modified rule of fixed number of neighbors
2023, Physica A: Statistical Mechanics and its ApplicationsExample-based large-scale marine scene authoring using Wang Cubes
2022, Visual InformaticsCitation Excerpt :Meanwhile, Vicsek et al. (1995) use the principles of statistical mechanics to calculate the forces and directions of individuals in a group. Since the Vicsek model can simulate complex systems with simple principles, researchers have applied it to the simulation of bird flocks other groups (Ginelli, 2016; Xueying et al., 2017; Chen et al., 2021). Besides the simulation of bird flocks, insect swarm simulation is also a trendy research area in group simulation.